Evidence
What we measure, and how.
What we measure, how we measure it, and the results so far.
Clinical Scribe: results
| Measure | How | Status |
|---|---|---|
| Scribe generation time | Timestamps in the app, per note, in real clinic use | 4.3s |
| Drug name accuracy | Compare the draft with the signed note, across 50,000+ Indian drug brands | 99.2% |
| Lines the doctor did not dictate | Count of marked lines, and whether the doctor kept or removed them | Measuring. Results will follow. |
| Consultations the model was trained on | Usage data across 200+ doctors | 500,000+ |
Each result will say how many notes, from how many doctors, between which dates, with omissions and additions reported separately. Small test sets will not be quoted as general accuracy.
Relaya Rx
Relaya Rx turns prescriptions into structured data with checks. It is not released. It stays unreleased until it passes our release gate, and we will publish the evaluation method before we publish any result.
How we report
Each figure here comes with the method used to measure it.